diff options
Diffstat (limited to 'fs/hfs/TODO')
| -rw-r--r-- | fs/hfs/TODO | 54 |
1 files changed, 54 insertions, 0 deletions
diff --git a/fs/hfs/TODO b/fs/hfs/TODO new file mode 100644 index 000000000..f989c3a91 --- /dev/null +++ b/fs/hfs/TODO @@ -0,0 +1,54 @@ +The hfs_fs "to do" list. +------------------------ +Items are broken down into groups and the groups are listed in order +from most important to least important. The items within each group +are not placed in any particular order. The order in which items are +listed probably doesn't correlate well with the order they will be +addressed. + +Genuine bugs: +1. Header files have compiled-in limit (currently 10) on descriptors. + +Missing features: +1. The partition code should be migrated into the kernel to allow + simultaneous access to multiple partitions on a single disk. +2. 1k block support is needed for some devices. +3. An ioctl()-based interface is needed to provide a consistent way + to do things under all of the representations of forked files. + +Possible additional "fork" mount options: +1. AppleSingle. +2. The scheme MacOS uses on FAT disks (PC Exchange). +3. "Flat" (no resource forks or metadata). + +Performance issues: +1. Use drAllocPtr to speed block allocations. +2. Keep a real cache of bnodes, rather than just a hash table of + the ones that are currently in use. +3. Keep a real cache of extent records, rather than just a linked + list of the ones that are currently in use and the one most + recently used. This is particularly needed to get acceptable + performance with multiple readers on a file. Perhaps simply + keep them in memory once they've been read until the file is + closed. + +Implementation details: +1. Allocation scheme could/should be closer to that used by Apple. +2. B*-tree insertion could/should be closer to that used by Apple. +3. Magic-number checks on data structures are rarely done. +4. Error recovery is needed for failed binsert(), bdelete() and rename(). +5. Deadlock detection is needed to make insert_empty_bnode() and + bdelete() less likely to hang on a corrupted B-tree. +6. Metadata for covered directories shouldn't appear in the filesystem. + Under CAP and AppleDouble it currently does. However, the obvious + solution is a real performance killer and is not worth implementing. + +Fantasy features: +1. Access Desktop file/database for comment and icon. +2. Implement mmap() for AppleDouble header files and CAP info files. +3. Implement AppleShare client support. + +Suggestions/comments/questions are welcome. +Code addressing any of the issues listed above is especially welcome. +Paul H. Hargrove +hargrove@sccm.Stanford.EDU |